Improving performance of NOMA and RSMA systems with improper Gaussian signaling

Abstract

With the new wave of beyond fifth generation (B5G) and sixth generation (6G) communication systems, there is a perpetual demand for more wireless services with higher data rates, lower latency, and greater connectivity. In order to meet these growing expectations, new candidate technologies (e.g., small cells, millimeter wave, and massive multiple-input multiple-output) have been introduced. Non-orthogonal multiple access (NOMA) has been presented among the most promising strategies for wireless applications due to its effectiveness in supporting heavily- loaded systems by serving users with diverse channel conditions in the same time-frequency resources. NOMA makes it possible to allocate one resource (frequency, time, code, or spatial) to serve multiple users at once by employing superposition coding at the transmitter side and successive interference cancellation (SIC) at the receiver side, resulting in more spectral-efficient and energy-efficient systems. Recently, rate-splitting multiple access (RSMA) has emerged as a more generalized multiple access technique than NOMA which can serve various under-loaded and over-loaded wireless applications by taking advantage of the common streams to better manage the interference. RSMA offers a flexible interference management technique by enabling an intelligent combination of transmitter-side and receiver-side interference mitigation rather than fully mitigating the interference at the receiver side as in NOMA. The RSMA strategy involves splitting user messages and employing a non-orthogonal transmission scheme, where common messages are decoded by multiple users, and private messages are decoded by their respective users. This approach enhances performance across a broader range of network loads, improving spectral and energy efficiency as well as user fairness. Improper Gaussian signalling (IGS) has emerged as a signal processing tool and potential alternative to the proper Gaussian signalling (PGS) to improve the spectral efficiency of interference-limited 5G and beyond networks. IGS achieves higher degrees of freedom than PGS due to its capability to control the interference signal dimension. IGS can be viewed as a type of interference alignment, where interference is effectively eliminated by confining it to a single orthogonal signal space dimension, allowing the desired signal to be decoded from the remaining orthogonal dimension. In this thesis, we investigate the potential performance merits of using IGS in the downlink interference-limited NOMA systems assuming practical scenarios including SIC imperfection in point-to-point NOMA systems and imperfect self-interference cancellation in cooperative full- duplex NOMA systems etc. We also investigate the potential performance merits of using IGS in RSMA as a generalization scheme of NOMA system. In the first part of this thesis, a point-to-point downlink NOMA system is studied, where the IGS strategy is adopted to compensate for the performance loss caused by imperfect SIC. New closed-form expressions for achievable user rates are derived when users employ the IGS strategy. Joint optimization problems are then formulated to maximize the overall spectral efficiency and energy efficiency of a two-user NOMA system, subject to minimum user-rate requirements and total power constraints. Sub-optimal solutions for the IGS circularity coefficients and power allocation are derived for the formulated problems. Additionally, improper constellation diagrams are designed using widely linear transformation (WLT) with the predefined optimized IGS coefficients to analyze the impact of IGS on throughput and error performance. In the second part of this thesis, a downlink cooperative full-duplex NOMA (FD-NOMA) system employing IGS under imperfect self-interference cancellation is analyzed. Optimization problems are formulated and solved to maximize the sum rate, achieve max-min rate fairness, and enhance energy efficiency. These problems involve the joint optimization of the circularity coefficients of the IGS and the power allocation at the base station, subject to each user’s rate constraints. We propose iterative algorithms based on solving the Karush-Kuhn-Tucker (KKT) conditions to derive sub-optimal solutions to the formulated problems. Additionally, we illustrate the impact of the IGS circularity coefficient on the constellation diagram of each user. In the third part of this thesis, we consider a downlink cellular system using RSMA transmission scheme at the base-station with IGS to serve multiple users. We first derive the achievable user private rate and user common rate considering IGS is used for the common message. Then, we maximize the private sum rate of the users’ private rates subject to certain minimum users’ common rate constraint. In this optimization problem, we optimize the IGS circularity coefficients and power allocation. The thesis results show that the performance of IGS-based NOMA/RSMA system outperforms its counterpart PGS-based NOMA/RSMA system under the realistic hardware imperfections.